← Models

Model profile

Smaug Llama 3 70B Instruct

2024-05-17release date
Not rankedSafety rank
#603 / 645Freedom rank

Evidence summary

Safety. Smaug Llama 3 70B Instruct does not meet the evidence gate for a Safety rank. Published Safety evidence spans 1 eval lineages and 3 of 7 components. Its strongest relative result is Enkrypt AI Safety Leaderboard (toxicity_attack_non_success_rate, #1 of 246); its weakest is Enkrypt AI Safety Leaderboard (harmful_attack_non_success_rate, #129 of 248).

Freedom. Smaug Llama 3 70B Instruct has an estimated Freedom rank of #603; its 90% source-sensitivity interval is #436–#624. Published Freedom evidence spans 1 eval lineages and 1 of 1 components. Its strongest relative result is Enkrypt AI Safety Leaderboard (harmful_attack_non_success_rate, #119 of 248); its weakest is Enkrypt AI Safety Leaderboard (toxicity_attack_non_success_rate, #242 of 246).

Compare this model

Only models sharing at least one published sub-eval are listed.

Official and reference links

Safety evals

Rank and direction are specific to the Safety portfolio. Black marks the observed result; the grey dot marks the value implied by the Safety rank. Values stay on each source’s native scale.

Eval / sub-evalRankValueDistributionSource
Enkrypt AI Safety Leaderboardbias_attack_non_success_rate#77 / 24822.22Source ↗official
Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate#56 / 24891.5Source ↗official
Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate#129 / 24869.44Source ↗official
Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate#1 / 246100Source ↗official

Freedom evals

Rank and direction are specific to the Freedom portfolio. Black marks the observed result; the grey dot marks the value implied by the Freedom rank. Values stay on each source’s native scale.

Eval / sub-evalRankValueDistributionSource
Enkrypt AI Safety Leaderboardcbrn_attack_non_success_rate#192 / 24891.5Source ↗official
Enkrypt AI Safety Leaderboardharmful_attack_non_success_rate#119 / 24869.44Source ↗official
Enkrypt AI Safety Leaderboardtoxicity_attack_non_success_rate#242 / 246100Source ↗official